1. Event Trees Quantitative Risk Analysis

2. Event Trees - Overview Definitions Steps Occurrence frequency Mean Time between Shutdown Mean Time Between Runaway Example

3. Accidents do happen! When an accident or process deviation (i.e. an “event”) occurs in a plant, various safety systems (both mechanical and human) come into play to prevent the accident from propagating. These safety systems either fail or succeed.

4. Event Trees Event trees are used to follow the potential course of events as the event moves through the various safety systems. The probability of success or failure of each safety intervention is used to determine the overall probability of each final outcome.

5. Event Trees An Event Tree is used to determine the frequency of occurrence of process shutdowns or runaway systems. Inductive approach Specify/Identify an initiating even and work forward. Identifies how a failure can occur and the probability of occurrence

6. Steps to Construct an Event Tree Identify an initiating event of interest. Identify the safety functions designed to deal with the initiation followed by the impact of the safety system Construct the event tree Describe the resulting accident event sequences.

7. Identify an initiating event May have been identified during a HAZOP as a potential event that could result in adverse consequences. Usually involves a major piece of operating equipment or processing step, i.e. a HAZOP “Study Node”.

8. Identify safety functions From PID, process flow sheet, or procedures find what safety systems are in place and what their functions are. These can include things such as automatic controllers, alarms, sensors, operator intervention, etc. On you Event Tree write across the top of the page in the sequence of the safety interventions that logically occur. Give each safety intervention an alphabetic letter notation.

9. Construct the Event Tree Horizontal lines are drawn between functions that apply Vertical lines are drawn at each safety function that applies Success – upward Failure – downward Indicate result of event Circle – acceptable result Cross-circle – unacceptable result

10. Construct Event Tree (cont.) Compute frequency of failures B is the failure per demand or the unavailability of safety function B

11. Occurrence Frequency Follow process through with each step to calculate the frequency of each consequence occurring. Typically three final results Continuous operation Shutdown (safely) Runaway or fail

12. Mean time between Shutdown Mean Time Between Shutdown, MTBS is calculated from: MTBS=1/  occurrences of shutdowns Mean Time Between Runaway, MTBR is calculated from: MTBR=1/  occurrences of runaways

13. Example – Loss of coolant to reactor Four safety interventions High temperature alarm Operator noticing the high temperature during normal inspection Operator re-establishes the coolant flow Operator performs emergency shutdown of reactor

14. Example – Loss of coolant Assume loss of coolant occurs once per year (occurrence frequency 1/yr) Alarm fails 1% of time placed in demand (failure rate of 0.01 failures/demand) Operator will notice high reactor temperature 3 out of 4 times (0.25 failures/demand) Operator will successfully restart coolant flow 3 out of 4 times (0.25 failures/demand) Operator successfully shuts down reactor 9 out of 10 times (0.10 failures/demand)

15. Resulting Event Tree Analysis

16. Example – Possible outcomes The lettering is used to identify each final outcome. For instance, ABDE Indicates that after Initiating event A occurs, that safety system B failed (high T alarm), that safety system D failed (the operator was unable to re-start the coolant) and safety system E failed (the operator was unable to successful shut down the reactor).

17. Example – Determination of MTBS For Mean Time Between Shutdowns take the reciprocal of the sum of all sequences that resulted in a shutdown. (Example gives 1/.225 = 4.4yrs) For Mean Time Between Runaway do the same thing with all sequences that resulted in a runaway. (Example gives 1/0.250 = 40yrs)

18. What is wrong with the logic of this example analysis?

19. What is wrong? If the operator fails to notice the high temperature after the alarms fails, then he/she will never restart the cooling.

20. In Class Example Construct an Event Tree and determine the MTBS and MTBR for a loss of coolant for the reactor shown in Figure 11-8. Assume loss of coolant occurs once every three years. Alarm fails 0.1% of time placed in demand Operator will notice high reactor temperature 3 out of 4 times Operator will successfully restart coolant flow 4 out of 5 times Operator successfully shuts down reactor 9 out of 10 times

21. Solution – Construct Event Tree

22. Solution Continued – Occurrence Frequency

23. Solution Continued – Mean Time Between Events